Accuracy needs an overhaul. Here's a realistic accuracy system.

[CenturionsofRome]

We all know that the accuracy system is in desperate need of an overhaul. Here's my idea of a system to replace it, which should be feasible from a programming point of view and accurate to history.

I think the best way to describe my proposed system is by example:

Ship X fires at Ship Y. Upon firing, the game calculates the flight time of X's shells and predicts where Y will be that many seconds in the future based on Y's current speed and heading. At that point (a literal point on the XY grid representing the predicted center of Y) the game will generate two dispersion circles. The first circle, called the Fire Control dispersion circle (FC Circle) represents the accuracy of X's FC systems and is centered on Y's predicted location and generates a point using a random dispersion algorithm (this generated point is where X thinks Y will be). The second dispersion circle is generated on the point created by the FC Circle, this is the Gun dispersion circle and all shells from X will land there. As the shells are physical objects in the game the game can track which compartment the shell hits, if it hits, and do the pen/damage calculations there.

Modifiers for the FC Circle:

• Base accuracy determined by Tower(s)
• First ranging shot (200% debuff)
• Second ranging shot (100% debuff)
• Shots on target (25% buff) [% probably needs adjustment]
• Under fire (40% debuff) [% probably needs adjustment]
• Weather (debuff ranging from mild to severe, reduced by radar)
• Technology modifier (Highly variable buff)
  • Primary FC destroyed (technology buff reduced by 1/3) [located in hull behind main belt]
  • Secondary FC destroyed (technology buff reduced by 2/3) [located in hull behind main belt] {some ships, like CLand DD, don't have a sec FC; in that case if the FC is hit both are considered destroyed}
• Rangefinders damaged (50% debuff) [Primary OR secondary tower destroyed]
• Rangefinders destroyed (100% debuff) [Primary AND secondary towers destroyed] {If ship has only one tower the rangefinders are destroyed with the tower}
• Conning Tower destroyed (20% debuff)
• Turning beyond 5 degrees resets ranging shot count (need to fire first and second ranging shots again)
• Ship obstructing view of target (50% debuff)
• Smoke obstructs target (200% debuff, negated by radar)
• Radar (150% buff) [Has a chance of being destroyed by battleship caliber guns (12" and higher) firing, the larger the gun the higher the chance. Gen 2 radar has reduced risk of being knockedd out by own guns firing]
  • Radar destroyed (radar bonuses removed) [caused by own guns firing, critical hit to main tower(?), or main tower destroyed]

Modifiers for Gun circle:

• Base accuracy determined by gun
• Ship instability (variable debuff derived from fore/aft and starboard/port offsets combined with hull base value) [battleships are naturally very stable, CLs and DDs are naturally quite unstable]
• Damage instability (variable debuff derived from amount of flooding damage)
• Sea state (debuff ranging from milld to severe)
• Turret(s) damaged (# of turrets damaged/100% debuff) [Turret indicator at least yellow, I.e if a ship has 2 turrets and one gets damaged it takes a 50% debuff. If both are damaged its a 100% debuff]
• Swamped (25% debuff) [turrets taking water due to low freeboard, possibly caused by flooding damage or very rough seas]

Notes: Each gun caliber has its own dispersion circle. Primary batteries and Secondary batteries have separate FC systems in order to allow for independent targeting.

Close readers will notice that its possible for ships to dodge incoming shells by altering course and/or speed after being fired on. That is both intentional and historically accurate, and it means that small, fast, and nimble ships, like CL and DD can dodge main gun batteries of capital ships at long range. However the shorter flight time and higher RoFs of smaller caliber guns, like 4", 5", and 6", are much harder to dodge. For those wondering why you'd need to make ranging shots again after turning, that's exactly what captains needed to do IRL after turning. 

Edited January 12, 2020 by CenturionsofRome

[Skeksis]

13 hours ago, CenturionsofRome said:

in desperate need of an overhaul...

Not really.

I've been trying to recreate some battles, started with Russo-Japanese War...

Battle of Port Arthur 1904

     Empire of Japan 

          6 pre-dreadnought battleships
          9 armoured cruisers
          10 destroyers

     Russian Empire

          7 pre-dreadnought battleships
          5 protected cruisers, with escorts

     Total: 37, 60 fps.

Battle of the Yellow Sea 1904

     Empire of Japan 
          4 Battleships
          4 Armoured Cruisers
          8 Protected Cruisers
          18 Destroyers.
          30 Torpedo Boats. 

     Russian Empire

          6 Battleships.
          4 Protected Cruisers.
          14 Destroyers.

      Total: 88, 5 fps (lowest).

Ok 88 ships is asking for abit too much but it's not unreasonable to ask to be an Admiral of a realistically large battle fleet of say about 25 ships. Best battle fleet would be about 2-4 BB/BC, 4-8 CA/CL, 8-16 DD/TB plus an equal sized enemy fleet. 

Is the campaign going to support this, I don't know but 30-50 ship battles could be on the cards after optimization (judging from custom battle capacity).

Why I'm pointing this out, if you want best or better realistic damage/accuracy model, there's going to be more calc's/algorithms for changes like "shells [to be] physical objects" and they are going suck-up so much more CPU crunching resources, then something else is going to have to be sacrifice at the other end, like less ships, less fires effects, less general effects, etc. I think we already loss about 15-20% ship capacity going from alpha II to III.

I believe the current gunnery model (with its abstract gunnery effects) allows us to be Admirals of a grand fleet and not captains of just a few ships.

Sorry but it's a no from me to replace the current model. 👎

Edited January 12, 2020 by Skeksis

[Teckelmaster]

I refer to one of my absolute all time favorite strategy games, when i answer this: Homeworld. I am actually old enough, that i played Homeworld when it originally was released. And Homeworkd had a system, where each projectile was „physically present“ and calculated. Homeworld 2 did not. The fanbase screamed in pain and agony, but to be perfectly honest, i did not notice a difference. I guess it is a lot harder to program a system calculating each individual projectile, rather than a simple rng thing. 
Since i am sure, that in terms of gaming experience i would not notice a huge difference, i do not really care too much about how they do it, as long as it works adequately enough. Here i refer to the experts in this forum, who know a lot about true accuracy, when i say, that currently it seems quite unhistorically and incorrect.
However, i mentioned in another thread, you can not hit the wrong ship accidentally, using this rng model. And since i play a lot, i had so many shells „land in wrong ships“, but simply falling through, that is a bit agonizing, to be perfectly honest. is it painfull enough to change a significant part of programming? I do not know. If devs say „sure, easy!“ then do it. It adds to realism, having true shells. If not, i do not have an oppinion. 
And finally i admit, that i do not understand the full depths of your proposed system, so i simply assumed the rest

[RAMJB]

Statistic-based gunnery model has been a staple of naval warfare games for a very long time. And has always given good results. Goes without saying that I'm in the "immersion/realism as a top priority" camp here, but even there calling for such a complex of "Buff-debuff" variables is not warranted when the baseline we have present already throws good results.

Now what I'm not that fond of is about some of the current gunnery modifiers as implemented. Some things don't scale with range as they should (ship size and speed can be a quite big issue for long range gunnery, it's not for point blank, for instance). Others don't seem present at all (even while firing to effect the longer you held a gunnery solution the better it got to be. Gunnery correction doesn't seem implemented thus far, at least not in the % modifiers). That kind of thing.

The gunnery model needs a good look on the way modifiers work, but not a revamp nor a drastic change. The baseline is more than good enough as is, and what's there only needs proper adjustment to better represent realistic gunnery - but there's no need for a complete overhaul and changeover of the system when what's there isn't broken at all, nor need to introduce a mind-boggling ammount of buff/debuff to replicate something that's easily achievable by simpler means.


And now I'm at it, no you couldn't "dogde" battleship shells. I know about the famous "chase the shot" story, but that was desperation at best and would only be tried in a scenario where an enemy would be firing at an individual ship. When you don't know who's shooting you and who's shooting someone else you can't "chase the shot". Not to mention what something like that would do to formation sailing. 
And I don't know where you get the idea that it would work on main guns but not on secondary guns "because their higher muzzle velocity", because it could be the case sometimes, it wouldn't on some others, and besides, smaller shells didn't hold speed well with distance while big main battery ones would reach much farther because of their much bigger inertia (they'd retain a higher speed up to much longer ranges), so effectively their time-on-target (or shell fight) time at secondary gun long ranges could very well (and usually was) noticeably larger than what the "slower" main caliber shells' would have.

Edited January 12, 2020 by RAMJB

[arkhangelsk]

My first thought is to remember this old thread. Your system looks a lot like his. The current accuracies in the game do require an overhaul (in fact, even reversion would be a huge improvement), but the consensus last time seems to be that the current abstraction system is OK.
https://forum.game-labs.net/topic/32671-the-gunnery-model/

The exact magnitudes aside, except for "Under Fire" and "Ship obstructing view of target" and "Swamped", most of the stuff you mention is already simulated in the modifiers. You can even lose the lock if you turn hard enough.

As for 5 degree turns, I'm not sure if reality is that strict (I've heard of back to back turns being a challenge, not little 5 degree turns). But even if it is, the game's point and click format means we can't enforce something this strict on the players. Unlike real ships, we don't have a navigator to compute an intercept solution and we have to guestimate even our gun arcs. Which means at least average skill players will have to issue course corrections a lot more frequently even if the target doesn't maneuver.

Your solution seems more like a halfway house with a lot of abstraction anyway (in that sense it's not much better than the game's current model). If you really want to play it right at all costs, then you will have to consider at least three error boxes. The first being present position error, this being a mix of bearing error and range error. On a perfectly stable ship, the bearing should be more or less perfect so various types of instability will be the main source of reading error. Range error will start with the rangefinder's inherent limits then add human error. Having decided how for present position deviated from true position, you'll have to consider course and speed measurement errors to finally generate the future position error, then overlay meteorological and gun errors ... etc, etc.

7 minutes ago, RAMJB said:

Statistic-based gunnery model has been a staple of naval warfare games for a very long time. And has always given good results. Goes without saying that I'm in the "immersion/realism as a top priority" camp here, but even there calling for such a complex of "Buff-debuff" variables is not warranted when the baseline we have present already throws good results.

I'll also point out one big edge of statistic-based gunnery models ... the relative ease of amendment. If this hotfix proved anything, it proves the relative ease that the assumptions underlying the game can be changed (in essence, they seem to have just entered massive numbers into the hit percentages of the little guns, jacked the M/V up and typed in unjustifiable numbers for penetration). It would be a lot harder to change (improve) anything if we have to think of CEP circles interacting with FC errors ... etc, etc.

Making adjustments easy causes improvements to become easy. It also can be evolved into customizability for the game. For example, for the production game, it may be possible to stuff the guns' basic characteristics into a CSV file. Players who really get a kick over this hotfix's super effective secondaries can just go ahead and put in massive numbers into the appropriate parts of the CSV file and BINGO, they can enjoy this hotfix all over again. Just make sure if you make Youtube videos out of it that you point this detail out, so people don't blame the game.

[DougToss]

Just out of curiosity, how do Jutland Pro and Rule the Waves 2 simulate gunnery? 

[Evil4Zerggin]

AFAIK RtW2 is fairly similar to UA:D, with a base chance based on range and multiplicative modifiers for maneuvering, target size, sea state, weather, smoke, etc. etc. though the specific multipliers are generally more sedate than UA:D's.

[RAMJB]

57 minutes ago, DougToss said:

Just out of curiosity, how do Jutland Pro and Rule the Waves 2 simulate gunnery? 

I'ts been ages since I last played Jutland, but if memory serves right, was based around statistic calculations aswell.

[disc]

11 hours ago, arkhangelsk said:

As for 5 degree turns, I'm not sure if reality is that strict (I've heard of back to back turns being a challenge, not little 5 degree turns).

The earliest computers could not deal very well with any substantial maneuvers. This is why the Germans' zig-zagging in WWI was so effective: The Royal Navy's Dreyer tables could not keep up, although I don't know how much of a turn they could deal with. Of course, the Germans couldn't get on target either, with their much more primitive systems.

Interwar, gunnery computers advanced enough to account for more own-maneuvers. A good example is the 1943 Battle of the Komandorski Islands. Both sides were able to shoot while freely maneuvering, but, crucially, they could not account for the other's maneuvers! After a point, the battle degenerated into the USS Pensacola chasing Japanese shots... then the Japanese kind of gave up and retreated.

 

French computers apparently were designed to account for radical enemy maneuvers by 1937. They used inclinometers to help account for changes in enemy course, though I am suspicious of their functionality at longer ranges.

British computers could not account well for enemy maneuvers until post WWII, with the Royal Navy's AFCT Mk10 on HMS Vanguard, their last battleship.

The Soviet Union's TsAS-0 computer would have been able to account for enemy maneuvers, but the battleships it would equip were never completed. I think the TsAS-1M, on the postwar-completed Chapayev cruisers, may have had this capacity. I do not know which computer the Sverdlovs used.

The US did not develop an equivalent system and moved on to guided weapons almost immediately after the war. 

The Japanese, Germans, and Italians did not either, to my knowledge.

[akd]

2 hours ago, disc said:

The US did not develop an equivalent system and moved on to guided weapons almost immediately after the war. 

?? Ford Rangekeeper computer.  Basic system with modifications stayed in use for controlling gunfire aboard Iowa until 1991.

Edited January 13, 2020 by akd

[disc]

I do not mean to imply the US, Germany, Italy, and Japan did not deploy fire-control computers.

 

The Ford Rangekeeper Mark 8, in service from about 1932, could not deal well with maneuvering targets, per Friedman's Naval Firepower. I rather suspect the French and postwar Soviet systems were not very good for this purpose, either, despite being built with that in mind. Likewise Royal Navy systems before the AFCT Mk10, which deployed postwar.

I don't think the US Computer Mark 1 could calculate maneuvers either. It was still the best naval AA computer in the world.

Edited January 13, 2020 by disc